skip named glyphs if code point already present in charmap

Author: mgieseki

src/FontEngine.cpp

@@ -169,29 +169,36 @@ void FontEngine::buildGidToCharCodeMap (RangeMap &charmap) const {
 	// In case the Unicode map of the font doesn't cover all characters, some
 	// of them could still be identified by their names if present in the font.
 	if (FT_HAS_GLYPH_NAMES(_currentFace)) {
+		NumericRanges<uint32_t> usedCodepoints;
+		for (size_t i=0; i < charmap.numRanges(); i++)
+			usedCodepoints.addRange(charmap.getRange(i).minval(), charmap.getRange(i).maxval());
 		if (charmap.empty())
-			addCharsByGlyphNames(1, getNumGlyphs(), charmap);
+			addCharsByGlyphNames(1, getNumGlyphs(), charmap, usedCodepoints);
 		else {
-			addCharsByGlyphNames(1, charmap.minKey()-1, charmap);
+			addCharsByGlyphNames(1, charmap.minKey()-1, charmap, usedCodepoints);
 			for (size_t i=1; i < charmap.numRanges(); i++)
-				addCharsByGlyphNames(charmap.getRange(i-1).max()+1, charmap.getRange(i).min()-1, charmap);
-			addCharsByGlyphNames(charmap.maxKey()+1, getNumGlyphs(), charmap);
+				addCharsByGlyphNames(charmap.getRange(i-1).max()+1, charmap.getRange(i).min()-1, charmap, usedCodepoints);
+			addCharsByGlyphNames(charmap.maxKey()+1, getNumGlyphs(), charmap, usedCodepoints);
 		}
 	}
 }
 
 
 /** Looks for known glyph names in a given GID range and adds the corresponding
- *  GID->code point mapping to a character map.
+ *  GID->code point mapping to a character map if the code point is not already present.
  *  @param[in] minGID minimum GID of range to iterate
  *  @param[in] maxGID maximum GID of range to iterate
- *  @param[in,out] charmap character map taking the new mappings */
-void FontEngine::addCharsByGlyphNames (uint32_t minGID, uint32_t maxGID, RangeMap &charmap) const {
+ *  @param[in,out] charmap character map taking the new mappings
+ *  @param[in,out] ucp collection of code points already present in the character map */
+void FontEngine::addCharsByGlyphNames (uint32_t minGID, uint32_t maxGID, RangeMap &charmap, CodepointRanges &ucp) const {
 	for (uint32_t gid=minGID; gid <= maxGID; gid++) {
 		char glyphName[64];
 		if (FT_Get_Glyph_Name(_currentFace, gid, glyphName, 64) == 0) {
-			if (int32_t cp = Unicode::aglNameToCodepoint(glyphName))
+			const int32_t cp = Unicode::aglNameToCodepoint(glyphName);
+			if (cp && !ucp.valueExists(cp)) {
 				charmap.addRange(gid, gid, cp);
+				ucp.addRange(cp);
+			}
 		}
 	}
 }

src/FontEngine.hpp

@@ -31,13 +31,15 @@
 #include "Character.hpp"
 #include "CharMapID.hpp"
 #include "Glyph.hpp"
+#include "NumericRanges.hpp"
 #include "RangeMap.hpp"
 
 class Font;
 
 /** This class provides methods to handle font files and font data.
  *  It's a wrapper for the Freetype font library. */
 class FontEngine {
+	using CodepointRanges = NumericRanges<uint32_t>;
 	public:
 		~FontEngine ();
 		static FontEngine& instance ();
@@ -75,7 +77,7 @@ class FontEngine {
 		FontEngine ();
 		bool setFont (const std::string &fname, int fontindex, const CharMapID &charmapID);
 		int charIndex (const Character &c) const;
-		void addCharsByGlyphNames (uint32_t minGID, uint32_t maxGID, RangeMap &charmap) const;
+		void addCharsByGlyphNames (uint32_t minGID, uint32_t maxGID, RangeMap &charmap, CodepointRanges &ucp) const;
 
 	private:
 		FT_Face _currentFace = nullptr;

src/ToUnicodeMap.cpp

@@ -36,17 +36,17 @@ bool ToUnicodeMap::addMissingMappings (uint32_t maxIndex) {
 	// collect Unicode points already in assigned
 	NumericRanges<uint32_t> codepoints;
 	for (size_t i=0; i < numRanges(); i++)
-		codepoints.addRange(rangeAt(i).minval(), rangeAt(i).maxval());
+		codepoints.addRange(getRange(i).minval(), getRange(i).maxval());
 	// fill unmapped ranges
 	bool success;
 	if (empty()) // no Unicode mapping present at all?
 		success = fillRange(1, maxIndex, 1, codepoints, true);
 	else {   // (partial) Unicode mapping present?
-		success = fillRange(1, rangeAt(0).min()-1, rangeAt(0).minval()-1, codepoints, false);
+		success = fillRange(1, getRange(0).min()-1, getRange(0).minval()-1, codepoints, false);
 		for (size_t i=0; i < numRanges()-1 && success; i++)
-			success = fillRange(rangeAt(i).max()+1, rangeAt(i+1).min()-1, rangeAt(i).maxval()+1, codepoints, true);
+			success = fillRange(getRange(i).max()+1, getRange(i+1).min()-1, getRange(i).maxval()+1, codepoints, true);
 		if (success)
-			success = fillRange(rangeAt(numRanges()-1).max()+1, maxIndex, rangeAt(numRanges()-1).maxval()+1, codepoints, true);
+			success = fillRange(getRange(numRanges()-1).max()+1, maxIndex, getRange(numRanges()-1).maxval()+1, codepoints, true);
 	}
 	return success;
 }

src/ToUnicodeMap.hpp

@@ -25,9 +25,8 @@
 #include "RangeMap.hpp"
 
 
-/** Represents a mapping from character indexes to unicode points. */
-class ToUnicodeMap : public RangeMap
-{
+/** Represents a mapping from character indexes to Unicode points. */
+class ToUnicodeMap : public RangeMap {
 	public:
 		bool addMissingMappings (uint32_t maxIndex);